Tick-borne encephalitis incidence forecasts for Austria, Germany, and Switzerland

Abstract

The forecast of human tick-borne encephalitis (TBE) incidence for the next years has been on the research agenda of epidemiologists since the discovery of this tick-borne zoonosis. Based on models to explain the trend as well as the low- and high-frequency oscillations in the Austrian TBE incidence series, TBE forecasts for Austria, Germany and Switzerland are presented here. For this purpose, generalized linear models (GLMs) of type negative binomial regression were calibrated with the TBE incidences of the period 1991–2018 to forecast the TBE incidences 2019 and 2020. The GLMs require only 4–5 predictors, 2 of which are large-scale synchronized over Central Europe and used for all 3 countries. Predictors used include the demographic parameters total population and net migration rate, the Scandinavian index which describes the large-scale atmospheric circulation patterns, the fructification index of the European beech (Fagus sylvatica) 2 years prior as a proxy for the intensity of the TBE virus transmission cycle, and the national TBE vaccination coverage. Since an official time series of TBE vaccination coverage is only available for Austria, the missing TBE vaccination coverages of Germany and Switzerland were reconstructed and presented as the first results. Model verification results in explained variances of 76% for Austria, 84% for Germany, and 89% for Switzerland. Thus, the best model fit was determined for the Swiss GLM which is able to predict the TBE incidence with a root-mean-square error of RMSE = 25 cases (19% of the mean TBE incidence 1991–2018 or 7% of the TBE incidence of 2018). Forecasting TBE incidences for 2019 and 2020 results in 92 ± 12 and 142 ± 26 TBE cases for Austria, 417 ± 71 and 670 ± 168 TBE cases for Germany as well as 235 ± 30 and 465 ± 91 TBE cases for Switzerland.